Why Cortiq shows you a plan before it does anything

Most assistants skip straight to the result. You ask, a paragraph appears, and the work that produced it is gone. When the task is small that is fine. When the task touches a wallet, a file or a decision someone else will rely on, you want to see the route as well as the destination.

Cortiq starts from the other end. Reason turns your request into numbered steps, and each step names one tool and the exact inputs it will use. If the planner found an address in your sentence, you see that address in the step. If a later step reuses an earlier result, you see the placeholder that wires them together.

Reading a plan takes seconds, and it catches the mistakes that matter: the wrong address, a step that should not be there, an action you did not ask for. You can fix any of it in Flow before pressing run.

The rule-based planner behind this is deliberately plain. It looks for concrete things in your words, such as addresses, transaction hashes, quoted text and numbers, and assembles fixed templates around them. The same words always give the same plan, which makes it easy to trust and easy to test. When a site operator enables the AI planner, its plans pass through the same checks: any tool that is not in the built-in list is refused.

None of this is meant to slow you down. It is meant to move the moment of review to where it is cheapest: before the work, not after.